The "sensor" in your link is from April the 1st, also known as the April fools day. The sensor is called Baceolus sensor - you might want to check out what it means in common English.
If it were real, the conversion would be noiseless. Generating RGB signals from continuous spectral data requires only weighting the photon count at each wavelength by the amplitudes of the tristimulus curves at that wavelength and then summing over all wavelengths for each of the three tristimulus curves. Noise is increased when the conversion from the sensor signal to a standard color space requires the use of negative numbers in the matrix. There are no negative numbers in the process required here.
That is one of the reasons a sensor like this would be nice to have. Hasn't happened yet.
Seems like this approach might be difficult with higher CRA applications...?
ReplyDeleteTalking about people that try to think outside the box in this kind of issues...
ReplyDeleteIs this just hot air? I'm sure it has downsides, I'd love to hear about them :)
"Wavelength-Detecting Sensor Eliminates Bayer Filter, Triples Resolution"
http://www.lensrentals.com/blog/2013/04/wavelength-detecting-sensor-eliminates-bayer-filter-triples-resolution
The downside is that it only works on April 1st for those who have faith.
DeleteThe "sensor" in your link is from April the 1st, also known as the April fools day. The sensor is called Baceolus sensor - you might want to check out what it means in common English.
DeleteTalking about April fools, there is another nice joke on Sony inventing a phase detection AF in every pixel:
Deletehttp://www.sonyalpharumors.com/sony-patent-discloses-new-on-sensor-phase-detection-af-technique/
Thanks to AA for the link!
I wonder how much noise the color conversion to RGB will create?
ReplyDeleteIf it were real, the conversion would be noiseless. Generating RGB signals from continuous spectral data requires only weighting the photon count at each wavelength by the amplitudes of the tristimulus curves at that wavelength and then summing over all wavelengths for each of the three tristimulus curves. Noise is increased when the conversion from the sensor signal to a standard color space requires the use of negative numbers in the matrix. There are no negative numbers in the process required here.
DeleteThat is one of the reasons a sensor like this would be nice to have. Hasn't happened yet.
The noise would be increased as long as the absolute of the number is larger than 1 during the color conversion, either positive or negative.
ReplyDelete